Abstract
In recent years Kansei retrieval systems have been developed to be practical use. We have developed the retrieval system using Kansei parameters based on online book reviews of children's books. There is a growing demand for Kansei retrieval systems that can retrieve the books whose value of Kansei parameters set high manually. Existing Kansei retrieval systems, however, cannot set appropriate value of Kansei parameters to those books featured by strong feelings, impressions or atmosphere, such as very amusing books or very sad. To resolve such problems, we experimented with various combinations of the methods to select the words in online book reviews and to set values of Kansei parameters in order to find the most appropriate combination to set parameters to those books. 1425 book reviews were used to assign the parameters automatically. The result showed that 1)the recall ratio was influenced considerably by the number of the words in book reviews, 2)the highest recall ratio was achieved when 200-300 words were used to assign parameters automatically, and 3)there was really not much difference between the approach of machine learning and multiple linear regression analysis.
